CtrlK
BlogDocsLog inGet started
Tessl Logo

the-provider-runtime-symmetry

Keep Stave's Claude and Codex provider runtimes in sync when editing one of them. Use when a change touches `electron/providers/claude-sdk-runtime.ts`, `electron/providers/codex-app-server-runtime.ts`, executable lookup, env construction, CLI session launch env, tooling status probes, or any provider-agnostic behavior. Trigger on phrases like "provider runtime", "claude adapter", "codex adapter", "sibling adapter", "PATH env", "CLAUDE_CONFIG_DIR", "CODEX_HOME", "tooling status", "어댑터 대칭", "양쪽 프로바이더 확인".

75

Quality

94%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

88%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A strong, highly actionable checklist skill: concrete file paths, real failure signatures, explicit validation steps, and a required output report. The main weaknesses are minor redundancy between the Guardrails section and the checklist, and a ~130-line body that could offload some sections to reference files for better progressive disclosure.

Suggestions

Trim the Guardrails section — its four bullets restate checklist items 1, 5, and 6; keep only the Zod/TS and centralize-vs-duplicate bullets that add new information.

Move 'SDK Upgrade Hygiene' and 'Common Failure Modes' into a reference file (e.g. references/upgrade-hygiene.md) and keep one-line summaries with clear links in SKILL.md, bringing the always-loaded body closer to a lean overview.

Convert inline doc path mentions (e.g. 'docs/developer/adding-a-provider.md') into clearly-signaled links under a 'References' heading so navigation to deeper material is unambiguous.

DimensionReasoningScore

Conciseness

The body is dense and project-specific — checklists, a file table, concrete failure signatures — with almost no explanation of concepts Claude already knows. However, the 'Guardrails' section largely restates checklist items ('Do not ship a fix in one adapter without verifying the sibling', 'Do not verify env changes with probes only' repeat sections 1 and 6), which is minor padding that could be trimmed. It sits between the level 5 'every token earns its place' anchor and level 4 'minor instances of over-explanation', closer to 4 due to this repetition.

4 / 5

Actionability

Guidance is fully executable for an instruction-only skill: exact file paths ('electron/main/ipc/schemas.ts', 'src/lib/providers/schemas.ts'), a concrete command ('bun run typecheck'), specific grep targets, exact failure output ('env: node: No such file or directory'), and precise verification surfaces (GUI launch, Settings → Providers). Matches the 'copy-paste ready ... specific examples cover the common cases' anchor; level 4's 'minor gaps' does not fit.

5 / 5

Workflow Clarity

The process is clearly sequenced: a seven-section symmetry checklist ordered by concern, followed by a verification section with explicit validation steps (typecheck always, real turns with each provider, GUI launch for env changes, probe-surface check for tooling-status changes) and a required output report. This matches the 'clear sequence with explicit validation steps; checklists for complex processes' anchor and exceeds level 4.

5 / 5

Progressive Disclosure

No bundle files exist (references/, scripts/, assets/ are absent), so the ~130-line body must stand alone; it is well organized with clear section headers, a file table, and inline pointers to repo docs ('docs/developer/adding-a-provider.md', 'docs/providers/codex-upgrade-checklist.md'). This is 'good structure; most content appropriately placed' (level 4) rather than 5: at this length, sections like 'SDK upgrade hygiene' and 'Common Failure Modes' could be split into reference files to slim the always-loaded body, and the doc pointers are inline code paths rather than clearly-signaled references.

4 / 5

Total

18

/

20

Passed

Description

100%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is exemplary: it states the what, an explicit 'Use when' clause with concrete file paths, and a comprehensive trigger phrase list including environment variable names and Korean variants. It occupies a clear niche with minimal conflict risk.

DimensionReasoningScore

Specificity

Concrete actions and coverage are explicit: 'Keep Stave's Claude and Codex provider runtimes in sync when editing one of them', 'executable lookup, env construction, CLI session launch env, tooling status probes, or any provider-agnostic behavior', plus exact file paths. Matches the 'multiple specific concrete actions; comprehensive coverage' anchor; nothing above it exists, and it is well beyond the 'several specific actions; minor gaps' level 4 anchor.

5 / 5

Completeness

Both what ('Keep Stave's Claude and Codex provider runtimes in sync when editing one of them') and when ('Use when a change touches ... Trigger on phrases like ...') are explicit with concrete trigger phrases, exactly matching the level 5 anchor. It clearly exceeds level 4, where the 'when' is present but less explicit.

5 / 5

Trigger Term Quality

Trigger coverage is comprehensive with synonyms and exact identifiers users would naturally say: 'provider runtime', 'claude adapter', 'codex adapter', 'sibling adapter', 'PATH env', 'CLAUDE_CONFIG_DIR', 'CODEX_HOME', 'tooling status', plus Korean variants. This matches the 'comprehensive coverage of natural terms including synonyms' anchor; level 4 ('a few natural terms missing') does not fit given both English and Korean variants and exact env-var names are present.

5 / 5

Distinctiveness Conflict Risk

The niche is a clear, narrowly-scoped domain (adapter parity between two named runtime files in the Stave codebase), so conflict risk with other skills is minimal. Level 4 ('minor overlap risk with closely related skills') does not apply because the description names the exact files and distinguishes itself from adjacent concerns like routing.

5 / 5

Total

20

/

20

Passed

Validation

93%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository
sendbird/stave
Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.